From: Guo Heyi <heyi.guo@linaro.org>
To: "Ni, Ruiyu" <ruiyu.ni@Intel.com>
Cc: Heyi Guo <heyi.guo@linaro.org>,
edk2-devel@lists.01.org, Yi Li <phoenix.liyi@huawei.com>,
Ard Biesheuvel <ard.biesheuvel@linaro.org>,
Star Zeng <star.zeng@intel.com>, Eric Dong <eric.dong@intel.com>,
Laszlo Ersek <lersek@redhat.com>,
Michael D Kinney <michael.d.kinney@intel.com>
Subject: Re: [PATCH v6 4/6] MdeModulePkg/PciHostBridgeDxe: Add support for address translation
Date: Thu, 15 Mar 2018 13:33:04 +0800 [thread overview]
Message-ID: <20180315053304.GB108227@SZX1000114654> (raw)
In-Reply-To: <26c26ca9-1d40-f206-ff17-7c445d5caefa@Intel.com>
On Thu, Mar 15, 2018 at 01:26:04PM +0800, Ni, Ruiyu wrote:
> On 3/15/2018 12:00 PM, Heyi Guo wrote:
> >PCI address translation is necessary for some non-x86 platforms. On
> >such platforms, address value (denoted as "device address" or "address
> >in PCI view") set to PCI BAR registers in configuration space might be
> >different from the address which is used by CPU to access the
> >registers in memory BAR or IO BAR spaces (denoted as "host address" or
> >"address in CPU view"). The difference between the two addresses is
> >called "Address Translation Offset" or simply "translation", and can
> >be represented by "Address Translation Offset" in ACPI QWORD Address
> >Space Descriptor (Offset 0x1E). However UEFI and ACPI differs on the
> >definitions of QWORD Address Space Descriptor, and we will follow UEFI
> >definition on UEFI protocols, such as PCI root bridge IO protocol and
> >PCI IO protocol. In UEFI 2.7, "Address Translation Offset" is "Offset
> >to apply to the Starting address to convert it to a PCI address". This
> >means:
> >
> >1. Translation = device address - host address.
> >
> >2. PciRootBridgeIo->Configuration should return CPU view address, as
> >well as PciIo->GetBarAttributes.
> >
> >Summary of addresses used in protocol interfaces and internal
> >implementations:
> >
> >1. *Only* the following protocol interfaces assume Address is Device
> > Address:
> >(1). PciHostBridgeResourceAllocation.GetProposedResources()
> > Otherwise PCI bus driver cannot set correct address into PCI
> > BARs.
> >(2). PciRootBridgeIo.Mem.Read() and PciRootBridgeIo.Mem.Write()
> >(3). PciRootBridgeIo.CopyMem()
> >UEFI and PI spec have clear statements for all other protocol
> >interfaces about the address type.
> >
> >2. Library interfaces and internal implementation:
> >(1). Base and Limit in PCI_ROOT_BRIDGE_APERTURE are device address.
> > It is easy to check whether the address is below 4G or above 4G.
> >(2). Addresses in PCI_ROOT_BRIDGE_INSTANCE.ResAllocNode are host
> > address, for they are allocated from GCD.
> >(3). Address passed to PciHostBridgeResourceConflict is host address,
> > for it comes from PCI_ROOT_BRIDGE_INSTANCE.ResAllocNode.
> >
> >RESTRICTION: to simplify the situation, we require the alignment of
> >Translation must be larger than any BAR alignment in the same root
> >bridge, so that resource allocation alignment can be applied to both
> >device address and host address.
> >
> >Contributed-under: TianoCore Contribution Agreement 1.1
> >Signed-off-by: Heyi Guo <heyi.guo@linaro.org>
> >Signed-off-by: Yi Li <phoenix.liyi@huawei.com>
> >Cc: Ruiyu Ni <ruiyu.ni@intel.com>
> >Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
> >Cc: Star Zeng <star.zeng@intel.com>
> >Cc: Eric Dong <eric.dong@intel.com>
> >Cc: Laszlo Ersek <lersek@redhat.com>
> >Cc: Michael D Kinney <michael.d.kinney@intel.com>
> >---
> >
> >Notes:
> > v6:
> > - Change ASSERT(FALSE) to ASSERT ((Translation & Alignment) == 0) and
> > move ASSERT before if-check [Ray]
> > - Print translation of bus and then assert [Ray]
> > - Add function header for RootBridgeIoGetMemTranslationByAddress()
> > [Ray]
> > v5:
> > - Add check for the alignment of Translation against the BAR alignment
> > [Ray]
> > - Keep coding style of comments consistent with the context [Ray]
> > - Comment change for Base in PCI_RES_NODE [Ray]
> > - Add macros of TO_HOST_ADDRESS and TO_DEVICE_ADDRESS for address type
> > conversion (After that we can also simply the comments about the
> > calculation [Ray]
> > - Add check for bus translation offset in CreateRootBridge(), making
> > sure it is zero, and unify code logic for all types of resource
> > after that [Ray]
> > - Use GetTranslationByResourceType() to simplify the code in
> > RootBridgeIoConfiguration() (also fix a bug in previous patch
> > version of missing a break after case TypePMem64) [Ray]
> > - Commit message refinement [Ray]
> > v4:
> > - Add ASSERT (FALSE) to default branch in GetTranslationByResourceType
> > [Laszlo]
> > - Fix bug when passing BaseAddress to gDS->AllocateIoSpace and
> > gDS->AllocateMemorySpace [Laszlo]
> > - Add comment for applying alignment to both device address and host
> > address, and add NOTE for the alignment requirement of Translation,
> > as well as in commit message [Laszlo][Ray]
> > - Improve indention for the code in CreateRootBridge [Laszlo]
> > - Improve comment for Translation in PCI_ROOT_BRIDGE_APERTURE
> > definition [Laszlo]
> > - Ignore translation of bus in CreateRootBridge
> > v4:
> > - Add ASSERT (FALSE) to default branch in GetTranslationByResourceType
> > [Laszlo]
> > - Fix bug when passing BaseAddress to gDS->AllocateIoSpace and
> > gDS->AllocateMemorySpace [Laszlo]
> > - Add comment for applying alignment to both device address and host
> > address, and add NOTE for the alignment requirement of Translation,
> > as well as in commit message [Laszlo][Ray]
> > - Improve indention for the code in CreateRootBridge [Laszlo]
> > - Improve comment for Translation in PCI_ROOT_BRIDGE_APERTURE
> > definition [Laszlo]
> > - Ignore translation of bus in CreateRootBridge
> >
> > MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.h | 21 +++
> > MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostResource.h | 3 +
> > MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.c | 129 ++++++++++++++++---
> > MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciRootBridgeIo.c | 135 ++++++++++++++++++--
> > 4 files changed, 261 insertions(+), 27 deletions(-)
> >
> >diff --git a/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.h b/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.h
> >index 9a8ca21f4819..c2791ea5c2a4 100644
> >--- a/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.h
> >+++ b/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.h
> >@@ -38,6 +38,13 @@ typedef struct {
> > #define PCI_HOST_BRIDGE_FROM_THIS(a) CR (a, PCI_HOST_BRIDGE_INSTANCE, ResAlloc, PCI_HOST_BRIDGE_SIGNATURE)
> > //
> >+// Macros to translate device address to host address and vice versa. According
> >+// to UEFI 2.7, device address = host address + translation offset.
> >+//
> >+#define TO_HOST_ADDRESS(DeviceAddress,TranslationOffset) ((DeviceAddress) - (TranslationOffset))
> >+#define TO_DEVICE_ADDRESS(HostAddress,TranslationOffset) ((HostAddress) + (TranslationOffset))
> >+
> >+//
> > // Driver Entry Point
> > //
> > /**
> >@@ -247,6 +254,20 @@ ResourceConflict (
> > IN PCI_HOST_BRIDGE_INSTANCE *HostBridge
> > );
> >+/**
> >+ This routine gets translation offset from a root bridge instance by resource type.
> >+
> >+ @param RootBridge The Root Bridge Instance for the resources.
> >+ @param ResourceType The Resource Type of the translation offset.
> >+
> >+ @retval The Translation Offset of the specified resource.
> >+**/
> >+UINT64
> >+GetTranslationByResourceType (
> >+ IN PCI_ROOT_BRIDGE_INSTANCE *RootBridge,
> >+ IN PCI_RESOURCE_TYPE ResourceType
> >+ );
> >+
> > extern EFI_METRONOME_ARCH_PROTOCOL *mMetronome;
> > extern EFI_CPU_IO2_PROTOCOL *mCpuIo;
> > #endif
> >diff --git a/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostResource.h b/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostResource.h
> >index 8612c0c3251b..a6c3739db368 100644
> >--- a/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostResource.h
> >+++ b/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostResource.h
> >@@ -38,6 +38,9 @@ typedef enum {
> > typedef struct {
> > PCI_RESOURCE_TYPE Type;
> >+ //
> >+ // Base is a host address
> >+ //
> > UINT64 Base;
> > UINT64 Length;
> > UINT64 Alignment;
> >diff --git a/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.c b/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.c
> >index 1494848c3e8c..5342efd89275 100644
> >--- a/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.c
> >+++ b/MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.c
> >@@ -33,6 +33,39 @@ EFI_EVENT mIoMmuEvent;
> > VOID *mIoMmuRegistration;
> > /**
> >+ This routine gets translation offset from a root bridge instance by resource type.
> >+
> >+ @param RootBridge The Root Bridge Instance for the resources.
> >+ @param ResourceType The Resource Type of the translation offset.
> >+
> >+ @retval The Translation Offset of the specified resource.
> >+**/
> >+UINT64
> >+GetTranslationByResourceType (
> >+ IN PCI_ROOT_BRIDGE_INSTANCE *RootBridge,
> >+ IN PCI_RESOURCE_TYPE ResourceType
> >+ )
> >+{
> >+ switch (ResourceType) {
> >+ case TypeIo:
> >+ return RootBridge->Io.Translation;
> >+ case TypeMem32:
> >+ return RootBridge->Mem.Translation;
> >+ case TypePMem32:
> >+ return RootBridge->PMem.Translation;
> >+ case TypeMem64:
> >+ return RootBridge->MemAbove4G.Translation;
> >+ case TypePMem64:
> >+ return RootBridge->PMemAbove4G.Translation;
> >+ case TypeBus:
> >+ return RootBridge->Bus.Translation;
> >+ default:
> >+ ASSERT (FALSE);
> >+ return 0;
> >+ }
> >+}
> >+
> >+/**
> > Ensure the compatibility of an IO space descriptor with the IO aperture.
> > The IO space descriptor can come from the GCD IO space map, or it can
> >@@ -366,6 +399,7 @@ InitializePciHostBridge (
> > UINTN MemApertureIndex;
> > BOOLEAN ResourceAssigned;
> > LIST_ENTRY *Link;
> >+ UINT64 HostAddress;
> > RootBridges = PciHostBridgeGetRootBridges (&RootBridgeCount);
> > if ((RootBridges == NULL) || (RootBridgeCount == 0)) {
> >@@ -411,8 +445,15 @@ InitializePciHostBridge (
> > }
> > if (RootBridges[Index].Io.Base <= RootBridges[Index].Io.Limit) {
> >+ //
> >+ // Base and Limit in PCI_ROOT_BRIDGE_APERTURE are device address.
> >+ // For GCD resource manipulation, we need to use host address.
> >+ //
> >+ HostAddress = TO_HOST_ADDRESS (RootBridges[Index].Io.Base,
> >+ RootBridges[Index].Io.Translation);
> >+
> > Status = AddIoSpace (
> >- RootBridges[Index].Io.Base,
> >+ HostAddress,
> > RootBridges[Index].Io.Limit - RootBridges[Index].Io.Base + 1
> > );
> > ASSERT_EFI_ERROR (Status);
> >@@ -422,7 +463,7 @@ InitializePciHostBridge (
> > EfiGcdIoTypeIo,
> > 0,
> > RootBridges[Index].Io.Limit - RootBridges[Index].Io.Base + 1,
> >- &RootBridges[Index].Io.Base,
> >+ &HostAddress,
> > gImageHandle,
> > NULL
> > );
> >@@ -443,14 +484,20 @@ InitializePciHostBridge (
> > for (MemApertureIndex = 0; MemApertureIndex < ARRAY_SIZE (MemApertures); MemApertureIndex++) {
> > if (MemApertures[MemApertureIndex]->Base <= MemApertures[MemApertureIndex]->Limit) {
> >+ //
> >+ // Base and Limit in PCI_ROOT_BRIDGE_APERTURE are device address.
> >+ // For GCD resource manipulation, we need to use host address.
> >+ //
> >+ HostAddress = TO_HOST_ADDRESS (MemApertures[MemApertureIndex]->Base,
> >+ MemApertures[MemApertureIndex]->Translation);
> > Status = AddMemoryMappedIoSpace (
> >- MemApertures[MemApertureIndex]->Base,
> >+ HostAddress,
> > MemApertures[MemApertureIndex]->Limit - MemApertures[MemApertureIndex]->Base + 1,
> > EFI_MEMORY_UC
> > );
> > ASSERT_EFI_ERROR (Status);
> > Status = gDS->SetMemorySpaceAttributes (
> >- MemApertures[MemApertureIndex]->Base,
> >+ HostAddress,
> > MemApertures[MemApertureIndex]->Limit - MemApertures[MemApertureIndex]->Base + 1,
> > EFI_MEMORY_UC
> > );
> >@@ -463,7 +510,7 @@ InitializePciHostBridge (
> > EfiGcdMemoryTypeMemoryMappedIo,
> > 0,
> > MemApertures[MemApertureIndex]->Limit - MemApertures[MemApertureIndex]->Base + 1,
> >- &MemApertures[MemApertureIndex]->Base,
> >+ &HostAddress,
> > gImageHandle,
> > NULL
> > );
> >@@ -654,6 +701,11 @@ AllocateResource (
> > if (BaseAddress < Limit) {
> > //
> > // Have to make sure Aligment is handled since we are doing direct address allocation
> >+ // Strictly speaking, alignment requirement should be applied to device
> >+ // address instead of host address which is used in GCD manipulation below,
> >+ // but as we restrict the alignment of Translation to be larger than any BAR
> >+ // alignment in the root bridge, we can simplify the situation and consider
> >+ // the same alignment requirement is also applied to host address.
> > //
> > BaseAddress = ALIGN_VALUE (BaseAddress, LShiftU64 (1, BitsOfAlignment));
> >@@ -721,6 +773,7 @@ NotifyPhase (
> > PCI_RESOURCE_TYPE Index2;
> > BOOLEAN ResNodeHandled[TypeMax];
> > UINT64 MaxAlignment;
> >+ UINT64 Translation;
> > HostBridge = PCI_HOST_BRIDGE_FROM_THIS (This);
> >@@ -822,14 +875,43 @@ NotifyPhase (
> > BitsOfAlignment = LowBitSet64 (Alignment + 1);
> > BaseAddress = MAX_UINT64;
> >+ //
> >+ // RESTRICTION: To simplify the situation, we require the alignment of
> >+ // Translation must be larger than any BAR alignment in the same root
> >+ // bridge, so that resource allocation alignment can be applied to
> >+ // both device address and host address.
> >+ //
> >+ Translation = GetTranslationByResourceType (RootBridge, Index);
> >+ ASSERT ((Translation & Alignment) == 0);
>
> Can you move the assertion to the body of if below, and below the DEBUG
> message? So that when assertion happens, developer can know the actual value
> of Translation and Alignment.
>
>
> >+ if ((Translation & Alignment) != 0) {
> >+ DEBUG ((DEBUG_ERROR, "[%a:%d] Translation %lx is not aligned to %lx!\n",
> >+ __FUNCTION__, __LINE__, Translation, Alignment
> >+ ));
>
> ASSERT ((Translation & Alignment) == 0); // Move to here.
No problem. I just found we often place ASSERT before if-check in other places
and supposed it is the common style :)
Thanks,
Heyi
>
> With the above change, Reviewed-by: Ruiyu Ni <ruiyu.ni@Intel.com>
>
> >+ //
> >+ // This may be caused by too large alignment or too small
> >+ // Translation; pick the 1st possibility and return out of resource,
> >+ // which can also go thru the same process for out of resource
> >+ // outside the loop.
> >+ //
> >+ ReturnStatus = EFI_OUT_OF_RESOURCES;
> >+ continue;
> >+ }
> >+
next prev parent reply other threads:[~2018-03-15 5:26 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-03-15 4:00 [PATCH v6 0/6] Add translation support to generic PciHostBridge Heyi Guo
2018-03-15 4:00 ` [PATCH v6 1/6] CorebootPayloadPkg/PciHostBridgeLib: clear aperture vars for (re)init Heyi Guo
2018-03-15 4:00 ` [PATCH v6 2/6] OvmfPkg/PciHostBridgeLib: clear PCI " Heyi Guo
2018-03-15 4:00 ` [PATCH v6 3/6] MdeModulePkg/PciHostBridgeLib.h: add address Translation Heyi Guo
2018-03-15 4:00 ` [PATCH v6 4/6] MdeModulePkg/PciHostBridgeDxe: Add support for address translation Heyi Guo
2018-03-15 5:26 ` Ni, Ruiyu
2018-03-15 5:33 ` Guo Heyi [this message]
2018-03-15 6:50 ` Ni, Ruiyu
2018-03-15 4:00 ` [PATCH v6 5/6] MdeModulePkg/PciBus: convert host address to device address Heyi Guo
2018-03-15 4:00 ` [PATCH v6 6/6] MdeModulePkg/PciBus: return CPU address for GetBarAttributes Heyi Guo
2018-03-15 5:27 ` [PATCH v6 0/6] Add translation support to generic PciHostBridge Ni, Ruiyu
2018-03-15 5:36 ` Guo Heyi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-list from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180315053304.GB108227@SZX1000114654 \
--to=devel@edk2.groups.io \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox